Solar Energy - The Built Environment Research Group

Page created by Lee Vasquez
 
CONTINUE READING
Solar Energy - The Built Environment Research Group
Solar Energy 218 (2021) 180–194

                                                                       Contents lists available at ScienceDirect

                                                                                  Solar Energy
                                                            journal homepage: www.elsevier.com/locate/solener

Optimal control of switchable ethylene-tetrafluoroethylene (ETFE)
cushions for building façades
Afshin Faramarzi, Brent Stephens, Mohammad Heidarinejad *
Department of Civil, Architectural, and Environmental Engineering, Illinois Institute of Technology, Chicago, IL, USA

A R T I C L E I N F O                                        A B S T R A C T

Keywords:                                                    Switchable ethylene-tetrafluoroethylene (ETFE) cushions with kinetic shading mechanisms are increasingly
Building energy performance                                  being used in building enclosures to dynamically control the transmission of solar and visible light. While
ETFE cushion, office building                                buildings with switchable ETFE façades typically utilize simple Rule-Based logic to control their operation, this
Optimal control
                                                             study uses a novel co-simulation approach to optimize the operation of switchable ETFE façades on two hypo­
Sequences of operation
                                                             thetical office buildings in Chicago, IL. Four seasonally representative simulation days are used to demonstrate
                                                             the approach. The daily source energy savings potential of the Optimal Control schedule is up to 8.2%, 11.1%,
                                                             and 25.5% compared to Rule-Based, Always-Dark, and Always-Bright control strategies, respectively.

                                                                                                    of a using completely delignified wood as a multifunctional, passive
                                                                                                    radiative cooling material composed of cellulose nanofiber bundles that
1. Introduction
                                                                                                    reflects solar radiation up to 96% (Li et al., 2019). Simulations indicated
                                                                                                    that an average of ~30% cooling energy savings can be obtained for
    The building sector is responsible for one third of global energy
                                                                                                    older midrise apartment buildings and an average of ~20% cooling
consumption (Berger & Mendes, 2017), with commercial and residential
                                                                                                    energy savings is feasible for new midrise apartments. There are also
building energy consumption expected to grow ~1.9% annually (Pérez-
                                                                                                    other studies that used intelligent control of renewable energy power
Lombard et al., 2008). The building enclosure, which includes the walls,
                                                                                                    generators applied for grid-interactive buildings (Mu et al., 2018, 2020).
roof, foundation, windows, and shading devices that separate the
                                                                                                        In addition to largely static building enclosure materials and tech­
conditioned interior environment from the exterior environment, im­
                                                                                                    nologies, architects and engineers have been designing, manufacturing,
pacts building energy consumption end-uses such as heating, cooling,
                                                                                                    and testing smart (i.e., dynamic) building enclosure technologies to
and lighting, as well as indoor environmental quality (IEQ) (Sadineni
                                                                                                    reduce thermal loads, provide thermal isolation, harvest energy, and/or
et al., 2011). Heat transfer through walls and windows alone are esti­
                                                                                                    improve IEQ (Menéndez et al., 2018). Dynamic building envelopes have
mated to account for a large proportion of heating and cooling energy
                                                                                                    the ability to modify their properties to meet energy efficiency goals,
needs in both residential (Huang et al., 1999) and commercial (Huang
                                                                                                    meet IEQ requirements, and interact with the electrical grid (DOE, 2019;
and Franconi, 1999) buildings.
                                                                                                    Tang et al., 2018). These dynamic building enclosure technologies
    A growing body of literature has demonstrated that novel envelope
                                                                                                    mostly operate under time-varying operation based on the interior and
technologies and materials can offer new opportunities to reduce cool­
                                                                                                    ambient conditions to achieve higher performance.
ing, heating, and/or lighting energy end-uses in buildings. For example,
                                                                                                        One promising trend in dynamic building envelope research is
one study demonstrated the use of passive strategies such as adding
                                                                                                    moving toward lightweight and transparent structures such as mem­
expanded polystyrene (EPS) thermal insulation, utilizing reflective
                                                                                                    brane and foil structures (Robinson-Gayle et al., 2001). Recent advances
coated glazing, adding overhangs, and white washing external walls
                                                                                                    in sensors, actuators, and control systems have made active adaptable
resulted in 31.4% energy savings and 36.8% peak load savings in a high-
                                                                                                    envelope systems applicable for use in buildings (Robinson-Gayle et al.,
rise building located in a hot and humid climate (Cheung et al., 2005).
                                                                                                    2001). These active systems are able to react instantly based on the
Another study showed that adding different thermal insulation to the
                                                                                                    feedback that they receive from indoor and ambient environmental
walls, roofs, and floor reduced total energy consumption by 20–40%,
                                                                                                    conditions. Consequently, these technologies create opportunities to
while reducing air infiltration resulted in another 20% of energy savings
                                                                                                    optimize thermal and optical performance of building envelopes and
(Chan and Chow, 1998). Another study demonstrated the effectiveness

 * Corresponding author.
   E-mail address: muh182@iit.edu (M. Heidarinejad).

https://doi.org/10.1016/j.solener.2021.01.059
Received 18 September 2020; Received in revised form 15 January 2021; Accepted 26 January 2021
Available online 11 March 2021
0038-092X/© 2021 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
Solar Energy - The Built Environment Research Group
A. Faramarzi et al.                                                                                                                  Solar Energy 218 (2021) 180–194

   Nomenclature                                                                      Variables
                                                                                     α           electricity site-to-source energy conversion factor
   Acronyms                                                                          β           gas site-to-source energy conversion factor
   BAS           Building Automation System                                          Lsch        zone lighting illuminance
   BPSO          Binary Particle Swarm Optimizer                                     Lz          zone lighting
   CSV           Comma Separated Value                                               N           total number of simulations
   EPA           The U.S. Environmental Protection Agency                            Qclg        cooling energy consumption
   ETFE          Ethylene-Tetrafluoroethylene                                        Qhtg        heating energy consumption
   EPS           Expanded Polystyrene                                                Qltg        lighting energy consumption
   IDF           Input Data File                                                         clg
                                                                                     Tsch        zone cooling temperature schedule
   IEQ           Indoor Environmental Quality                                            htg
   LBNL          Lawrence Berkeley National Laboratory                               Tsch        zone heating temperature schedule
   NIST          National Institute of Standards and Technology                      Tz          zone temperature
   SHGC          Solar Heat Gain Coefficient
                                                                                     →u          control trajectory
   SOO           Sequences of Operation                                              J           Joule
   TMY           Typical Meteorological Year                                         lx          lux (luminous flux)
   TOU           Time-of-Use
                                                                                     ◦
                                                                                       C         degree of centigrade
   VCBT          Virtual Cybernetic Building Testbed

enhance the energy efficiency of buildings (Biloria and Sumini, 2009;                environmental forces such as wind. By controlling the air pressure be­
Flor et al., 2018).                                                                  tween cavities of different ETFE foil layers, the position of the middle
    Ethylene-tetrafluoroethylene (ETFE) is one of the façade technolo­               layer can be adjusted, for example by switching between dark mode
gies that has gained interest due to its light weight and high daylight              (Fig. 2a) to bright mode (Fig. 2b) and vice versa. Since the layers are
transmittance (Poirazis et al., 2009). When ETFE foils (sheets) are used             printed with inks and material additives, providing the ability to reflect/
on building façades, they are assembled into inflatable cushions. ETFE               transmit light with different shapes and arrangements applied in
has comparable or even better optical, thermal, mechanical, and struc­               different layers of the ETFE foil, the dark and bright positions yield
tural characteristics compared to glass structures. It is also lighter than          different thermal and optical properties for the façade. It is noted that
glass. These key features together render ETFE popular among architects              this pneumatically controlled mechanisms, which is commonly referred
and engineers in recent years (Hu et al., 2017). A few prominent                     to as ‘switchable’ ETFE cushions, is a relatively new technology applied
buildings worldwide, such as the Eden project in the UK, the Allianz                 in only a small number of buildings worldwide (Flor et al., 2018).
Arena football stadium in Germany, the National Aquatics Center                          A growing body of studies has assessed the mechanical and structural
(Fig. 1a), the Changzhou Flora Expo in China (Fig. 1b), and the Kaplan               performance of ETFE foils, for example, evaluating mechanical behavior
Institute building in the U.S. (Fig. 1c) have utilized ETFE in their fa­             of ETFE foils under uniaxial monotonic tension (Hu et al., 2017; Zhao
çades. Moreover, several prominent buildings worldwide have used                     et al., 2020), under tensile loading conditions (Charbonneau et al.,
switchable ETFE cushions with a kinetic shading mechanism that can be                2014), and during the form-developing process (Zhao et al., 2016). A
employed in roof and wall systems to dynamically control the trans­                  limited number of studies have assessed the performance of ETFE
mission of solar and visible light, with the goals of saving energy,                 cushions in terms of reducing building energy end-uses and/or
reducing peak power demands, and improving IEQ.                                      improving daylighting, including a few studies that have used building
                                                                                     energy modeling to assess the performance of ETFE installed on building
2. Literature review on ETFE façades                                                 envelopes. For example, Cremers and Marx (Cremers and Marx, 2016)
                                                                                     conducted a comparative analysis on new infrared (IR)-absorbing ETFE
    ETFE foils applied to buildings commonly consist of different con­               film with conventional silver printing ETFE foil for a hypothetical
figurations, ranging from just one layer to multiple layers (Flor et al.,            building with a dimension of 20 m width, 10 m length, and 5 m height
2018). In multiple layer installations, multiple layers (i.e., ranging from          with a total volume of 1,000 m3 located in Stuttgart, Germany. They
2 to 5) of ETFE foils are assembled to create a cushion with air pockets in          estimated that the cooling energy and heating energy savings are 5–8%
between the layers. The cushions are then connected to air compressor                and 6–7%, respectively. Cremers and Marx conducted another study
(s) through a pipe network, which acts as a pneumatic system that sta­               (Cremers and Marx, 2017) to evaluate the energy performance of
bilizes the air pressure. The air pressure is usually between 300 and 600            spatially transformed ETFE-foil (3D-foil). In a 3-D foil, the printing
Pa and causes the inflated ETFE cushions to withstand external                       pattern is adjusted exactly to the sun position in the sky. The cooling

Fig. 1. ETFE structures: (a) National Aquatics Center (Cushion ETFE), (b) Changzhou Flora Expo (single layer ETFE) (adapted from (Hu et al., 2017)), and (c) Kaplan
Institute building (four layers switchable ETFE cushion).

                                                                               181
Solar Energy - The Built Environment Research Group
A. Faramarzi et al.                                                                                                               Solar Energy 218 (2021) 180–194

                      Fig. 2. (a) Dark and (b) Bright mode of switchable ETFE with three layers (figure adapted from Flor et al., 2018).

energy savings of a new 3D foil compared to a simple one-layer ETFE foil           example, in order to maximize the effects of the ETFE cushion and to
for the same building as the previous study was estimated as 69%. In               more accurately represent real installations, we use a 95% window-to-
order to have the most savings, the pattern in 3D foils must be printed            wall ratio for the case study buildings. To reduce computational run
based on the sun position in each project, which presents a limitation to          time, we consider only 5 thermal zones. The building schedules and
the practical use of 3D foils. In their case, the optimal printing pattern         other construction properties are modeled as the ASHRAE 189.1-2009
was estimated by the sun position in Stuttgart, Germany.                           High-Performance buildings standard available in the OpenStudio li­
    Flor et al. (2018) investigated the energy savings potential of a              brary (ASHRAE, 2009).
building with different frit patterns printed on switchable ETFE cushions              The building energy model simulation engine is EnergyPlus V.9.1.0
and compared it to traditional clear and reflective glazing. They esti­            for optimization purposes (Ellis et al., 2008). The initial modeling was
mated an annual energy savings potential (for heating, lighting and                accomplished by OpenStudio (V2.8.0). Fig. 3 shows the schematic view
cooling end-uses) of about 45% for a switchable ETFE cushion compared              of these buildings. The small size building is considered as core-
to reflective double-glazing material. They also estimated that 65%                perimeter zone with 1 core thermal zone and 4 perimeter thermal
more hours of useful daylight can be obtained in spaces covered by ETFE            zones with in total 5 thermal zones. It has one floor with an area of
compared to reflective glazing. Afrin developed a simple simulation                508.7 m2 with a core zone area of 107.4 m2 and each perimeter zone
model for the heat transfer through the ETFE structures and then vali­             area of 100.33 m2. The model is considered as square with equal width
dated and calibrated the model by comparing it with measured data                  and length of 22.6 m along with 3.05 m height. The depth of perimeter
from test rigs and a case study building (Afrin, 2016).                            zones in each model is considered as 6.1 m. The zones are separated with
    While these existing studies show promise for energy savings po­               internal partitions, with the lines shown on the roof of the building in
tential, and ETFE cushion façades have already been built in several               Fig. 3. The HVAC system is considered as ideal air load system with dual
locations globally, buildings with switchable ETFE façades typically               thermostat setpoint schedule for the small office building located in
utilize fairly simple Rule-Based logic to control their operation. ASHRAE          Chicago, IL. The heating temperature setpoint is set to 21 ◦ C from 6 am
Guideline 36, which aims to summarize a list of best practices for the             to 10 pm during occupied times with a temperature setback of 15.6 ◦ C
sequences of operation (SOO) for buildings, does not currently include             for unoccupied times. The cooling temperature setpoint is set to 24 ◦ C
any recommendations for these façade types (ASHRAE, 2018). To the                  from 6 am to 10 pm during occupied times with a setback of 26.7 ◦ C for
best of our knowledge, there have not been any studies that have                   unoccupied times. For daylighting control, the setpoint is fixed at 500
explored optimal operation or control strategies for switching ETFE fa­            lux for each zone, referenced by a node. As seen in Fig. 3, the original
çades. Therefore, this study seeks to explore optimal dynamic control              models are revised with the exterior side of perimeter zones considered
strategies for actively interacting with the interior and ambient envi­            as ETFE by modeling it as window in the initial design. All the thermo-
ronments and varying the thermal and optical performance of ETFE                   physical properties and HVAC system configuration for the medium size
cushion façades to reduce building energy and improve daylighting in               office building (Fig. 3a) are the same as the small office building
hypothetical case study buildings. We use a co-simulation approach to              (Fig. 3b), except the geometry. The medium size office has three floors,
optimize the operation of ETFE façades for two hypothetical office                 with each floor of 1,618.7 m2, including a core zone with floor area of
buildings in Chicago, IL USA using a set of practical constraints.                 786.3 m2 and four perimeter zone with the floor area of 208.1 m2 for
                                                                                   each zone. This layout results in a square shape building with a total
3. Methodology                                                                     floor area of 4,856.1 m2 with three floors.
                                                                                       We assume the building uses the Texlon® “Vario” ETFE system,
    The problem considered here is an optimal control problem, dis­                which is a three-layered pressure inflated panel. The interior and exte­
cretized in time. The problem is defined to find the control trajectory of         rior layers have a pattern printed on their inward facing sides, and dark
ETFE status in three building directions (east, west, and south).                  and bright conditions are adjusted by pushing the printed middle layer
Considering the discretized optimal control problem, taking into ac­               toward the interior and exterior layers. Based on the definition from the
count each control node (time) as design variables, along with assuming            manufacturer, Vector-foiltec, the print pattern combinations are SQM
two switching ETFE status (i.e., either dark or bright) as the domain of           200-197:45 dark (Maywald, 2019). Table 1 shows the optical and
design variables, we convert the subject to a non-linear binary optimi­            thermal properties of this system under the bright and dark modes, taken
zation problem. For solving this problem, we employed an advanced                  from (Maywald, 2019). Relevant thermal and optical properties include
version of swarm intelligence binary optimizer (BPSO) with V-Shape                 the transmittance, reflectance, and absorptance for ultraviolet (UV),
transfer function.                                                                 visible, and solar light, shown separately for the bright and dark modes
                                                                                   of the ETFE cushion. Visible properties provide information needed to
                                                                                   design lighting fixtures while solar properties assist designers in esti­
3.1. Case study building energy models
                                                                                   mating the potential impact of the façade heating and cooling loads.
                                                                                   Table 1 also presents thermal properties including the Solar Heat Gain
   Two hypothetical office buildings are considered in this study. Their
                                                                                   Coefficient (SHGC), which characterizes the amount of heat gain ach­
areas are similar to U.S. Department of Energy (DOE) Reference build­
                                                                                   ieved by solar radiation through the ETFE cushion, the Shading Coeffi­
ings (Deru et al., 2011) for small and medium size office buildings.
                                                                                   cient (SC), which characterizes the amount of solar heat gain of a
However, they differ from the reference buildings in other ways. For

                                                                             182
Solar Energy - The Built Environment Research Group
A. Faramarzi et al.                                                                                                                      Solar Energy 218 (2021) 180–194

                              Fig. 3. Schematic view of the modified DOE Reference office building models: (a) medium size and (b) small size.

                                                                                          30) and three solstice and equinox days (March, June, and December 21)
Table 1
                                                                                          are selected for simulation. The majority of the analysis and results are
Thermal and optical properties of the Bright and Dark modes of switchable ETFE
                                                                                          conducted on the shoulder season day (September 30) to demonstrate
cushions.
                                                                                          the approach because shoulder seasons provide the most opportunity for
  Texlon® Vario       Bright                           Dark                               optimal control of dynamic façades due to larger diurnal fluctuations in
                      UV          Visible    Solar     UV      Visible     Solar          ambient conditions than solstice or equinox days. Only day-long simu­
                      light       light      light     light   light       light          lations are utilized because of the computational time required for
  Transmittance       15%         26%        26%       5%      9%          10%            annual analysis and because day-long simulations provide clearer
  Reflectance         14%         40%        38%       14%     53%         52%            insight into dynamic operational characteristics.
  Absorptance         71%         34%        35%       81%     38%         39%
  g-value/SHGC        0.37                             0.14
  SC                  0.42                             0.16                               3.3. Optimization algorithm
  U-value (W/         2.78                             2.78
    m2-K)                                                                                     There are many problems that intrinsically have the nature of
                                                                                          discrete binary search space. The problem such as dimensionally
                                                                                          reduction and feature selection can be categorized as binary problems
specific fenestration material compared to the standard clear float glass,
                                                                                          (Pal and Maiti, 2010; Zeng et al., 2009). The problem with continuous
and the U-value, which characterizes the heat transfer transmittance of
                                                                                          search space can also be converted to a binary problem type witch has
the material.
                                                                                          own its structure with some limitation (Mirjalili and Lewis, 2013). A
                                                                                          binary search space can be imagined as a hypercube in which the search
3.2. Control strategies                                                                   agents are only allowed to shift to the close/far corners of this hypercube
                                                                                          with selecting various number of bits (Kennedy and Eberhart, 1997). It is
    We consider four control strategies in this study: (1) Always-Dark,                   possible to design the binary version of originally continuous optimi­
(2) Always-Bright, (3) Rule-Based, and (4) Optimal Control. The ETFE                      zation method (Emary et al., 2016; Kennedy and Eberhart, 1997). The
operation is assumed to be the same for all floors facing the same di­                    continuous optimization methods are revised in structure with some­
rection. For example, for the east side, all the ETFE in three floors of the              times considering proper transfer function to map the search agent up­
medium size office building are assumed to be controlled by one                           dates into the domain dealing with only two numbers of “0” and “1”. In
schedule and controller. This applies for the south and west sides as well,               Particle Swarm Optimization (PSO), the transfer function is responsible
although different facing façades are assumed to operate independently.                   to connect a link between position and velocity of the particles. The
The north side is not connected to any controller and is assumed to be                    velocity values are converted to probability values for updating the
always bright due to the lack of direct sunlight in the studied latitudes.                positions of the particles. The traditional PSO uses sigmoid function as
For both building case studies, the SOO for each control strategy is as                   transfer function of velocities into probabilities. Since the sigmoid
follows:                                                                                  transfer function is like the word “S”, a group of sigmoid functions with
                                                                                          shape of “S” are called an S-shaped family of transfer functions (Mirjalili
 • Always-Dark: ETFE actuators are always acting, which results in                        and Lewis, 2013). Another family of transfer functions, the V-shaped
   constant dark status regardless of the time of day, occupancy, or                      family, have also shown better performance than the S-shaped family
   orientation of the space;                                                              (Mirjalili and Lewis, 2013). One of the recently developed functions of
 • Always-Bright: ETFE actuators are never acting, which results in                       V-shape family has shown more promising performance compared to
   constant bright status. Similar to the Always-Dark strategy, the time                  others in this class (Mirjalili and Lewis, 2013). The reason for better
   of day, occupancy, and orientation of the spaces do not impact ETFE                    performance of this class of functions is because of its equal behavior
   operation;                                                                             (resulting in equal probability) with status updating of the design vari­
 • Rule-Based: ETFE actuators act when the outdoor air temperature is                     ables that have the same absolute velocity. Since the optimal control
   above 15.6 ◦ C. This rule is inspired by the SOO that is currently being               problem of ETFE cushion envelope is a binary problem, we used the
   applied in the Kaplan Institute building (Fig. 1c). Therefore, we                      binary version of PSO with V-shape transfer function for the optimiza­
   consider this as a simple Rule-Based control to be compared with the                   tion purpose.
   Optimal Control. The temperature of 15.6 ◦ C is based on the building
   automation system (BAS) parts of architectural/construction draw­                      3.4. Co-simulation architecture
   ing of the building with considering a few simplifications; and
 • Optimal Control: ETFE actuators will act based on the optimal                              We used a “static” co-simulation architecture to couple the optimizer
   schedule derived from minimization of total daily heating, cooling,                    (MATLAB) and simulator (EnergyPlus). In a static co-simulation, data is
   and lighting energy consumption.                                                       exchanged just one time between the clients at each iteration/timestep
                                                                                          (Zhai, 2003). In this work, the optimizer (MATLAB) is responsible for
    In order to evaluate and compare the performance of these control                     writing the optimal schedule of ETFE status into a CSV (Comma-Sepa­
strategies during different days, a day in a shoulder season (September                   rated Value) file in an iteration, which serves as the input for the

                                                                                    183
A. Faramarzi et al.                                                                                                                    Solar Energy 218 (2021) 180–194

simulator (EnergyPlus). Conversely, the simulator (EnergyPlus) is                   properties and used a switchable glazing control type to switch between
responsible for writing the results of heating, cooling, and lighting loads         ETFE conditions according to the external schedule file in a CSV format
as inputs for the optimizer (MATLAB) in an iteration. Consequently, the             indicating ETFE operation mode with binary numbers (0 and 1). The
input of one client works as the output for the other client. The data are          switching mechanism is first carried out by assigning the dark status in
exchanged at each timestep (which in this case is each iteration). To               “window control shading” and bright status in “construction” tabs in the
achieve this, a syntax in MATLAB called “system” is used to call Ener­              IDF editor section on EnergyPlus. Then the binary number of 0 and 1 in
gyPlus. This syntax calls the operating system to execute a specified               schedule CSV file is used to actuate the status as either “off” (as bright
command. The defined command here to execute EnergyPlus entails a                   mode) or “on” (as dark mode), respectively. Interested readers can visit
character vector that includes three strings. The first string is the               the EnergyPlus Input-Output Reference Manual for more details (Ener­
directory address of the executable EnergyPlus file in the command                  gyPlus, 2000).
prompt called “RunEplus”. The second string includes the IDF infor­
mation that contains the name of the IDF file along with its directory.
The third string entails the weather data information (i.e., file name and          3.5. Objective function and design variables
its directory). The “system” will return the “status” and “cmdout” of the
execution. The “status” indicates if the command completed successfully                 The objective function considered in this study is the total daily
or not and the “cmdout” returns the output of the command. Fig. 4                   source energy use for heating, cooling, and lighting. Cooling and light­
shows the architecture of the data exchange in more detail.                         ing both use electricity and heating uses natural gas. Site to source en­
    As mentioned, we used OpenStudio for the initial modeling of our                ergy conversion factors are considered as 3 for electricity, as an average
case study; however, OpenStudio in the co-simulation architecture is                of value derived from the National Renewable Energy Laboratory
never actually called and run. After revising the geometry and assigning            (NREL) (Deru and Torcellini, 2007) and the US Environmental Protec­
HVAC systems and construction materials in OpenStudio, we run the                   tion Agency (EPA), and 1.05 for natural gas per the US EPA’s conversion
initial simulation from OpenStudio to generate the EnergyPlus input                 table. The design variables (optimal control nodes) are defined as the
data file (IDF). Then, the IDF file is employed in the proposed co-                 ETFE status in each of the east, south, and west directions. The execution
simulation architecture. However, before using the IDF file, we need                horizon for the optimal control problem is considered from 5 am to 8
to modify it and prepare it for data exchange ability. The modification             pm, with two control nodes in each hour and each direction of east,
steps are as follows:                                                               south, and west. In other words, we assume that the ETFE status can be
                                                                                    switched every 30 min, which results in 30 control nodes in each di­
 • Schedule type limit: defined the limit of ETFE operation status as               rection between 5 am to 8 pm, and 90 control node/design variables for
   discrete numeric type define by 0 and 1;                                         the problem in total. Thus, the optimal control of ETFE is mathemati­
 • Construction: Defined two status of ETFE as dark and bright mode;                cally formulated as Equation (1):
 • Window material: Assigned optical and thermal properties of the                                                   ∑N (                  ) ∑N
                                                                                       Minimize J(→ u E, →
                                                                                                         u S, →
                                                                                                              u W) =        α Qclgi + Qltgi +          βQhtgi
   dark and bright ETFE status previously defined in “Construction”;                                                    i=1                        i=1
                                                                                                  ⎧
 • Daylighting reference point: Defined the reference point for                                   ⎪                                        ⎧
                                                                                                  ⎪→
                                                                                                  ⎪               30
   daylighting control in each zone;                                                              ⎨ u E ∈ {0, 1}
                                                                                                  ⎪                                                  htg
                                                                                                                                           ⎨ Tz ∈ [Tsch      clg
                                                                                                                                                         , Tsch  ] (1)
 • Daylighting control: Assigned the reference point defined in previous             Considering
                                                                                                  ⎪
                                                                                                     →u S ∈ {0, 1}30 , and subject to
                                                                                                                                           ⎩ L ∈ [L ]
                                                                                                  ⎪
   section along illuminance setpoint for daylighting control;                                    ⎪
                                                                                                  ⎩→
                                                                                                  ⎪  u W ∈ {0, 1}30
                                                                                                                                                 z       sch

 • Schedule-file: defined the schedule file name for each orientation of
   east, south, and west; and
 • Window shading control: Assigned the shading type as “Switchable                 where → u is the control trajectory, Nis the number of total simulation
   Glazing” and shading control type as “on if schedule allows”.                    time steps in a specified day (the time step set 10 min as the EnergyPlus
                                                                                    default recommendation), α and β are the site-to-source energy con­
   Within EnergyPlus, we modeled the ETFE dark and bright conditions                version factors for electricity and heating, respectively, Qclg , Qltg and Qhtg
simply as two different window types with different optical/thermal                 are the site cooling, lighting, and heating energy consumption (J),
                                                                                    respectively, Tz is the zone temperature (˚C), and Lz is the zone lighting
                                                                                    illuminance (lux). Zone temperatures and zone lighting illuminance are
                                                                                    both constraints for the minimization problems.

                                                                                    4. Results

                                                                                        Detailed results from the four simulated control strategies are
                                                                                    explored for a single day representative of a typical shoulder season in
                                                                                    Chicago, IL (September 30). A shoulder season timeframe is considered
                                                                                    because it captures days when the outdoor temperature is both above
                                                                                    and below the outdoor temperature assumed to trigger action in the
                                                                                    Rule-Based control strategy (15.6 ◦ C), which allows for comparison
                                                                                    among all four of the simulated strategies. For example, if during a
                                                                                    summer day the outdoor air temperature fluctuates from 20 ◦ C to 30 ◦ C,
                                                                                    two of the comparison control strategies (Rule-Based and Always-Dark)
                                                                                    would yield the same results.
                                                                                        Fig. 5 shows the outdoor dry bulb temperature, direct solar radiation
                                                                                    flux, and sky clearness from the typical meteorological year (TMY3)
                                                                                    weather file on the simulated day (EnergyPlus, 2020). Fig. 5a shows that
                                                                                    the maximum outdoor air temperature is around 19 ◦ C at 1:00 pm while
                                                                                    the minimum is 8 ◦ C at 10:00 pm. Fig. 5b shows the sun rises at 6:00 am
                                                                                    and sets at 5:30 pm, with two sudden decreases around 7:30 am and
      Fig. 4. Co-simulation architecture of simulator-optimizer platform.           11:30 am due to cloud cover (Fig. 5c).

                                                                              184
A. Faramarzi et al.                                                                                                                    Solar Energy 218 (2021) 180–194

               Fig. 5. Outdoor conditions on the simulated day in September: (a) outdoor air temperature, (b) solar irradiation, and (c) sky clearness.

4.1. Small office building                                                               requires additional lighting energy earlier in the day than the south and
                                                                                         west zones, which is consistent with solar geometry patterns at this
    The ETFE variable considered in this study is binary, meaning that                   latitude. The south-facing perimeter zone has the highest cooling energy
the ETFE system can switch between either of two extreme conditions:                     consumption due to more solar gain that it receives during the entire day
dark or bright. Fig. 6 shows the assumed schedules for Always-Bright                     (Fig. 7b). Fig. 7c shows that heating energy consumption for all zones
operation (Fig. 6a), Always-Dark operation (Fig. 6b), and Rule-Based                     are more or less the same, occurring only in the early morning and
control operation (Fig. 6c). The Rule-Based control strategy is applied                  nighttime in this shoulder season day due to decreases in outdoor tem­
to all directions at the same time; when the outdoor temperature in­                     perature, as shown in Fig. 5a.
creases beyond 15.6 ◦ C (e.g., from 10:30 am to 5:00 pm on the simu­                         Fig. 8 shows lighting, cooling, and heating energy end-uses in the
lation day, as shown in Fig. 5a), the ETFE cushion switches to dark status               Always-Dark mode of operation during the September simulation day.
for all sides. This simple Rule-Based control is directly informed by                    As shown in Fig. 8a, lighting energy consumption in Always-Dark mode
discussions with facility managers at the Kaplan Institute building who                  is higher than in the Always-Bright mode, especially on the east, west,
operate an ETFE cushion façade.                                                          and south perimeter zones. Lighting energy consumption for the east
    Fig. 7 shows lighting, cooling, and heating energy end-uses in the                   and west zones intuitively intersect at the middle of the day, while the
Always-Bright mode of operation during the September simulation day.                     south zone is lowest. Fig. 8b shows that there is no cooling required for
Lighting energy use in the core zone is higher than the other zones, with                the east, north, or core zones at any time, while the cooling energy
a peak energy consumption around 550,000 J during a 10-minute                            consumption is considerably lower than the Always-Bright mode for the
simulation interval from 8:00 am to 5:00 pm. Since the ETFE is always                    south and west side zones. Fig. 8c shows that the heating energy con­
bright, all the thermal zones on the perimeter need approximately the                    sumption required in the east, west, and south zones is similar to, albeit
same pattern of lighting energy, with a minimum energy consumption                       a little higher than, the Always-Bright mode. The reason for minor dif­
around 150,000 J during a 10-minute interval within that period. The                     ferences is because heating needs are only from 6:00 am to 9:30 am and
reason that the perimeter zones still need lighting with the ETFE cushion                6:00 pm to 10:00 pm, when solar radiation is minimal.
operating in the Always-Bright mode is that in order to control the light,                   Fig. 9 shows lighting, cooling, and heating energy consumption for
we used continuous dimming control in EnergyPlus. Based on the                           the Rule-Based control strategy applied uniformly to all façades on the
default recommendation of EnergyPlus, we set the minimum input                           September simulation day. Referring back to Fig. 6c, from 10:30 am to
power fraction (the lowest power the lighting system can dim down to,                    5:00 pm, the ETFE façade is in dark mode operation (with ambient
expressed as a fraction of maximum input power) to 0.2 and the mini­                     temperatures above 15.6 ◦ C), and in bright mode the rest of the time.
mum light output fraction (the lowest lighting output the lighting sys­                  Consequently, Fig. 9a shows that the lighting energy use in each zone in
tem can dim down to, expressed as a fraction of maximum light output)                    Rule-Based control operation follows the same trend of the lighting load
to 0.3. Lighting energy use increases for all perimeter zones starting                   in Always-Dark mode from 10:30 am to 5:00 pm, while in other hours it
around 2:30 pm because of a decrease in solar light due to the overcast                  follows the trend of Always-Bright mode. Fig. 9b shows the cooling
sky at that time, as observed in Fig. 5b and c. Moreover, the east side                  energy consumption for this control strategy. As expected, the cooling

                      Fig. 6. Schedule of ETFE for (a) Always-Bright, (b) Always-Dark, and (c) Rule-Based operation on the September simulation day.

                                                                                   185
A. Faramarzi et al.                                                                                                                     Solar Energy 218 (2021) 180–194

Fig. 7. Site energy consumption of the small office building using the Always-Bright strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

Fig. 8. Site energy consumption of the small office building using the Always-Dark strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

Fig. 9. Site energy consumption of the small office building using the Rule-Based control strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

energy pattern is similar to the cooling energy pattern of Always-Dark                 south, and west ETFE cushion façades in the small office building during
mode operation (Fig. 8b), although the cooling load on the south-                      the September simulation day. Resulting time-varying estimates of
facing perimeter zone is slightly larger than Always-Dark because                      lighting, cooling, and heating energy consumption under the Optimal
before 10:30 am, the ETFE is in bright mode, which allows transmitted                  Control strategy on the same simulation day are shown in Fig. 11. In the
solar radiation to heat the south zone, resulting in more cooling required             east perimeter zone, the ETFE façade is operated in bright mode from 30
for the zone subsequently. Fig. 9c also shows that heating energy con­                 min after sunrise (6:30 am) until 9:30 am to benefit from both
sumption in each zone in the Rule-Based control strategy is similar to the             daylighting and solar heating. From 9:30 am to 10:30 am, the east ETFE
Always-Bright mode, as heating is only required when Rule-Based con­                   façade changes to dark status, when the optimizer determines that
trol calls only for coincident bright mode.                                            changing to dark status for a one-hour period would increase lighting
    Fig. 10 shows the resulting Optimal Control schedule for the east,                 needs but decrease cooling needs to a greater extent. From 10:30 am

                                                                                 186
A. Faramarzi et al.                                                                                                                     Solar Energy 218 (2021) 180–194

Fig. 10. Optimal Control schedule of ETFE status for three façades on the small office building during the September simulation day: (a) east, (b) south, and (c) west.

Fig. 11. Site energy consumption of the small office building using the Optimal Control strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

until sunset at 5:30 pm, the east ETFE façade remains in bright mode                   4.2. Medium office building
operation to benefit from daylighting without intense solar radiation.
After 5:30 pm, the ETFE façade changes again to dark mode. We also                         We considered the same core-perimeter zone concept for the medium
noticed that if the schedule would remain as bright after 5:30 pm, there               size office building. In order to reduce computational run time, all three
would be a slight increase in heating energy because the ambient tem­                  floors on each side were considered as one thermal zone. Therefore, we
perature keeps decreasing after 5:00 pm and keeping the ETFE in bright                 again have one core zone along with four perimeter zones, even in this
mode would affect the office by radiative heat transfer to the outside.                larger case study. The core zone has a larger area and volume compared
    In the south perimeter zone, the ETFE façade is operated in bright                 to perimeter zones. The lighting, cooling, and heating energy con­
mode from sunrise until 9:00 am to take advantage of natural                           sumption for the two base strategies of Always-Bright and Always-Dark
daylighting, then it switches to dark status until 3:00 pm to decrease                 during the September simulation day are presented in Fig. 12 and
cooling energy needs. From 3:00 pm to 5:30 pm, the south façade again                  Fig. 13, respectively.
switches to bright status because solar radiation is not intense in the                    Based on Fig. 12a, the lighting energy consumption of the core zone
evening, so the cooling demand is not large, while lighting energy is                  is considerably higher than the other zones. This is because this zone is
minimized. For the west perimeter zone, the ETFE façade is operated in                 not only larger than the other zones, but also because internal partitions
bright mode from sunrise until 12:00 pm. This duration is longer than                  that encircle the core zone prevent this zone from benefitting from any
the east and south faces because the west side does not face direct                    natural daylighting. Therefore, changing the status of ETFE will not have
sunlight until afternoon. Therefore, the optimizer takes the most                      any effect on the lighting of core zone of medium office. It also applies
advantage of daylighting before noon without increasing the cooling                    for the small office, but in the small office, the core zone is not much
energy beyond lighting saves. Subsequently, from 12:00 pm to 4:00 pm,                  larger than the perimeter zones, so the lighting energy required for the
the west ETFE façade changes to dark status in order to minimize cooling               core zone in the medium size office is considerably higher than the
energy consumption (the peak cooling demand when the ETFE is in                        lighting energy of the core zone needed in small office.
bright mode occurs at 2:40 pm, as shown in Fig. 7b). The west side ETFE                    Fig. 12b presents the cooling energy consumption of each zone under
façade then switches to bright mode from 4:00 pm to 5:30 pm, as solar                  Always-Bright operation. It is interesting to note that even though the
radiation decreases, and the benefits of natural daylighting are greater               core zone in the medium size office is larger than the other zones, the
than the impacts on cooling energy needs in that timeframe. Finally,                   cooling energy consumption of the core and south-facing zones are
after 5:30 pm, the ETFE cushion switches to dark status on all sides.                  competitive with each other due to high solar gains on the south side
Overall, the net impact of each control strategy on daily energy con­                  during the day. East, north and west side perimeter zones have lower
sumption is summarized in Section 4.3.                                                 cooling energy than the core and south side zones. Fig. 12(c) shows the
                                                                                       heating energy for this mode of operation. In contrast to lighting and
                                                                                       cooling, the core zone requires the lowest heating energy for this model

                                                                                 187
A. Faramarzi et al.                                                                                                                  Solar Energy 218 (2021) 180–194

Fig. 12. Site energy consumption of the medium office building using the Always-Bright strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

Fig. 13. Site energy consumption of the medium office building using the Always-Dark strategy during the September simulation day: (a) lighting, (b) cooling, and
(c) heating.

because it is surrounded and thermally protected by the perimeter                    substantially lower than in the Always-Bright mode. Fig. 13(c) shows
zones, with heat gains from people and equipment contributing to                     that the heating energy use for all zones in Always-Bright mode are only
meeting heating needs. In this larger case study building, heating is only           slightly higher than in Always-Dark mode, for reasons similar to those
required during short periods of time in the morning and early night to              explained for the small office building.
satisfy the temperature setpoints.                                                       Fig. 14 shows the resulting optimal schedule for the east, south, and
    Similarly, Fig. 13a depicts lighting energy use for the Always-Dark              west ETFE cushion façades in the medium size office building during the
mode of operation during the September simulation day. The west,                     September simulation day. The schedule for Always-Dark, Always-
south, and east zones have higher lighting energy use in Always-Dark                 Bright and Rule-Based control are the same as Fig. 6. Fig. 15 presents
mode compared to the same zones in Always-Bright mode, while light­                  lighting, cooling, and heating energy consumption for the Optimal
ing energy use in the north and core zones are the same as in Always-                Control strategy during the same simulation day.
Bright mode. Fig. 13b shows the cooling energy in the Always-Dark                        Based on Fig. 14a, the ETFE façade on the east perimeter zone is in
mode, where the loads in all zones except the core zone are                          bright mode until 8:00 am to decrease heating energy and lighting

Fig. 14. Optimal Control schedule of ETFE status for three façades on the medium office building during the September simulation day: (a) east, (b) south, and
(c) west.

                                                                               188
A. Faramarzi et al.                                                                                                                         Solar Energy 218 (2021) 180–194

Fig. 15. Site energy consumption of the medium office building using the Optimal Control strategy during the September simulation day: (a) lighting, (b) cooling,
and (c) heating.

energy needs in the morning, then it switches to dark mode until 1:30
                                                                                       Table 3
pm. Based on Fig. 12b, if ETFE is in bright mode, the east zone requires
                                                                                       Total daily source energy savings from using the Optimal Control strategy
cooling from 9:20 am until 4:00 pm. Therefore, this switch will decrease
                                                                                       compared to the three other strategies in the small and medium size office
the cooling energy needs while increase the lighting energy needs, but in
                                                                                       buildings during the September simulation day.
order to have lowest energy consumption, the optimizer identifies
                                                                                        Total daily source energy savings (%)
greater benefits in decreasing cooling energy needs. From 1:30 pm to
5:30 pm, the east ETFE façade is in bright mode to minimize lighting                    Optimal Control savings compared to the other    Small size      Medium size
energy, which has greater benefits than decreasing concurrent cooling                   strategies                                       office          office

energy needs. For the south side, the ETFE façade is operated in bright                 vs. Always-Bright                                25.5%           21.9%
mode from sunrise until 7:00 am to benefit lighting needs, then switches                vs. Always-Dark                                  11.1%           2.2%
                                                                                        vs. Rule-Based                                   8.2%            3.7%
to dark status until 4:00 pm (near sunset), and then switches to bright
until 6:30 pm. The reasons for these switches are the same as those
discussed previously for the small office building. The west side opera­               respectively, for the small size office building. Similarly, the Optimal
tion is also similar to the small office building as well, albeit with shorter         Control strategy resulted in total savings of 21.9%, 2.2%, and 3.7%
duration and a shift in hours of operation.                                            compared to Always-Bright, Always-Dark and Rule-Based control,
    One of the major differences between the optimal ETFE schedule for                 respectively, for the medium size office building. Intuitively, the reason
the medium size office building and the small office building is that in               that the energy savings potential of the Optimal Control strategy applied
the medium size office, the schedule in all directions tends to switch to              to the medium size office is lower than the small size office is because of
the dark mode during most hours of the day. The reason is the sensitivity              the greater contribution of the core zone in the larger building, which is
of the zones’ cooling energy (except the core zone) to the different ETFE              not affected by ETFE position.
status (Fig. 12b and Fig. 13b), which results in the optimizer giving
priority to cooling energy needs as the key load for optimization. We
also observed that the cooling and lighting energy consumption of the                  4.4. Other solstice and equinox days
core zone is much higher than the other zones and does not change with
different ETFE operational modes.                                                          In order to evaluate the effect of the ETFE Optimal Control strategy
                                                                                       on building energy consumption patterns during other seasons, we ran
4.3. Energy consumption and savings                                                    simulations for three additional simulation days, including solstice days
                                                                                       of June 21 and December 21, and the equinox day of March 21, each as
    Table 2 shows the estimated total daily source energy consumption                  representative days for summer, winter, and spring shoulder seasons,
for both building case studies for the September simulation day under                  respectively. It is noted that we already discussed the detailed results of
each of the four ETFE cushion control strategies.                                      September 30, which is close to another equinox day of September 23.
    Table 3 shows the percentage source energy savings resulting from
the use of the optimal ETFE cushion control strategy compared to the                   Table 4
three other strategies for the September simulation day. The Optimal                   Total daily source energy consumption for the small and medium size office
                                                                                       buildings under each of four ETFE cushion control strategies for the 21st of
Control strategy resulted in total savings of 25.5%, 11.1%, and 8.2%
                                                                                       December, June, and March.
compared to Always-Bright, Always-Dark, and Rule-Based control,
                                                                                        Total daily source energy consumption (J)

Table 2                                                                                 Control         Small size office                  Medium size office
Total daily source energy consumption for small and medium size office build­           Strategies
                                                                                                        Dec 21     June         March      Dec 21     June 21    March
ings during the September simulation day under each of four ETFE cushion                                           21           21                               21
control strategies.
                                                                                        Always-         1.38   ×   2.03     ×   9.69 ×     9.25   ×   2.20 ×     7.17 ×
  Total daily source energy consumption, J (kWh/m2)                                       Bright        109        109          108        109        1010       109
                                                                                        Always-Dark     1.48   ×   1.56     ×   1.02 ×     9.77   ×   1.80 ×     7.27 ×
  Control Strategies           Small size office           Medium size office
                                                                                                        109        109          109        109        1010       109
  Always-Bright                7.21 ×   108 (0.393)        8.07 × 109 (0.450)           Rule-Based      1.38   ×   1.56     ×   9.69 ×     9.25   ×   1.80 ×     7.17 ×
  Always-Dark                  6.04 ×   108 (0.329)        6.44 × 109 (0.359)                           109        109          108        109        1010       109
  Rule-Based                   5.85 ×   108 (0.319)        6.54 × 109 (0.365)           Optimal         1.38   ×   1.53     ×   8.73 ×     9.25   ×   1.78 ×     6.50 ×
  Optimal                      5.37 ×   108 (0.292)        6.30 × 109 (0.351)                           109        109          108        109        1010       109

                                                                                 189
A. Faramarzi et al.                                                                                                                     Solar Energy 218 (2021) 180–194

This section briefly describes and analyzes the outcomes of these days,                  up to 19% at the highest threshold temperature. The reason can be
albeit with less detail than the September day for brevity. Table 4 shows                explained in Fig. 12 and Fig. 13, which show that the lighting energy
the total daily source energy consumption estimated for the small and                    consumption of the medium office core zone is considerably higher than
medium size office buildings for these days and Table 5 presents the                     the other zones due to its larger area. Consequently, switching the ETFE
estimated energy savings of the Optimal Control strategy compared to                     status from Always-Bright to Always-Dark does not change the total
the three other control strategies during these same days.                               lighting energy consumption much, and while the heating energy con­
    The Optimal, Always-Bright, and Rule-Based control strategies all                    sumption is also not very sensitive to Always-Bright or Always-Dark
reached the same energy consumption on December 21 because the                           modes, the cooling energy consumption is more sensitive to the mode
continuously cold outside temperature causes the ETFE cushions switch                    of operation. Thus, the cooling energy will be the most sensitive factor
to the Always-Bright condition at all times to take the most advantage of                guiding the direction of problem. It is clear from Fig. 13 that in Always-
sunlight for both daylight harvesting and heating. Similarly, on June 21,                Dark mode, the cooling energy of the perimeter zones are significantly
the Optimal, Always-Dark, and Rule-Based control strategies all reached                  lower compared to the Bright mode. This is why the lower ambient
to nearly the same results because with continuously high outside                        temperature threshold of 12.6 ◦ C is the optimum threshold temperature
temperatures, the ETFE tends to switch to Always-Dark mode to                            compared to the default of 15.6 ◦ C. Also, based on Table 2, for the
decrease the cooling energy consumption as much as possible. Results                     medium office, the Always-Dark mode has lower energy consumption
from the Optimal Control strategy were slightly better than the Rule-                    compared to the Rule-Based Control with 15.6 ◦ C as the threshold
Based and Always-Dark control strategies: approximately 2% and 1%                        temperature. It is noted that the threshold temperature would most
lower energy consumption for the small office and medium size office                     likely be different for different office sizes, and the threshold tempera­
buildings, respectively. Greater magnitudes of savings for the Optimal                   ture needs to be tuned for buildings with different envelope and ge­
Control strategy are observed on March 21, which as a spring shoulder                    ometry characteristics. Additionally, similar to the small office building,
season day is closer to the September simulation day in that diurnal                     the energy consumption for all temperature thresholds remains the same
variations in ambient conditions provide greater potential for                           for the other simulation days of December 21, June 21 and March 21.
optimization.                                                                            The result of this temperature threshold value analysis further illustrates
                                                                                         that sometimes the Rule-Based Control or sequences of operation that
4.5. Impacts of different temperature threshold values for the rule-based                are currently being applied in the industry perform worse than the
control                                                                                  optimal (or near optimal) control.
                                                                                             Table 7 summarizes the total daily source energy consumption for
    This section evaluates the impacts of different ambient temperature                  the medium size office building during the September simulation day for
threshold values to switch the ETFE modes from dark mode to bright                       default and optimum thresholds based on the results of Table 6. Table 8
mode or vice versa in the Rule-Based Control strategy. An ambient                        shows the associated daily savings for the two switch temperature values
temperature range of 11.6 ◦ C to 18.6 ◦ C with step size of 1˚C was                      shown in Table 7. Based on Table 8, the amount of energy savings po­
considered to assess the impacts of the switch temperature in addition to                tential of the Optimal Control strategy compared to the Rule-Based
the originally considered 15.6 ◦ C. Table 6 presents the results for esti­               Control strategy with default and optimum threshold temperatures in
mated total daily source energy consumption for the small and medium                     the medium office would be 3.7% and 1.9%, respectively.
size office building for all four simulation days considered earlier and
under the different temperature threshold values for the Rule-Based                      4.6. Economic analysis
Control strategy.
    For the small office building, in Sep 30 as the representative of a                      This section provides a brief assessment of the costs of installing and
shoulder season day, the minimum energy consumption occurs at a                          operating ETFE building façades compared to conventional construc­
threshold of 15.6 ◦ C (5.85 × 108 J). It is observed that with a deviation               tion, which is intended to provide some insight into the relative costs
from this threshold temperature in either direction, the energy con­                     and benefits of this type of system.
sumption on this simulation day would increase, for example by up to
3% for the lowest temperature thresholds and by up to 18% for the                        4.6.1. Cost of ETFE inflating/switching action
highest threshold. For the other simulation days of December 21, June                        This section estimates the annual energy cost of the inflating/
21 and March 21, the energy consumption for all temperature thresholds                   switching action to change the ETFE status in the hypothetical small and
remains the same for this control strategy. The reason is that the tem­                  medium size office buildings in Chicago, IL under assumptions of con­
perature variation throughout in these days is consistently either above                 stant utility rates and Time-of-Use (TOU) utility rates. Table 9 shows the
or below the thresholds defined in Table 6, which results in no dynamic                  utility rate of constant and TOU programs for different Peak, Off-Peak
changes of the ETFE status throughout the representative simulation                      and Super-Peak daily period (ComEd, n.d.).
days.                                                                                        The inflating action is done by constantly blowing air from the
    Conversely, for the medium size office building, the lowest energy                   compressor into the ETFE chambers and the switching action is carried
consumption is achieved for the Rule-Based Control on September 30                       out by varying air flow and adjusting the air pressure in each cushion. By
with a switch mode temperature of 12.6 ◦ C (6.42 × 109 J), increasing by                 this adjustment, the internal layer changes their position inside the
                                                                                         cushion and the status changes. Therefore, the annual energy cost for the
Table 5                                                                                  inflating/switching action is estimated by expanding the ETFE inflating/
Total daily source energy savings from using the Optimal Control strategy                switching cost of representative simulation days to represent their entire
compared to the three other strategies in the small and medium size office               associated seasons and summing up the costs in each season to reach the
buildings for the 21st of December, June, and March.                                     annual cost. For the shoulder season, we used the daily cost of inflating/
  Total daily source energy savings (%)
                                                                                         switching the ETFE status on the September 30 simulation day, which
                                                                                         was already achieved through the Optimal Control schedule, and
  Optimal Control         Small size office           Medium size office
                                                                                         assumed this day replicated during each day of the shoulder season
  savings compared to
  the other strategies
                          Dec      June       March   Dec      June        March         months including April, May, September, and October. This scenario
                          21       21         21      21       21          21
                                                                                         was also used for the representative day of the winter season, December
  vs. Always-Bright       0.0%     24.6%      9.9%    0.0%     19.1%       9.3%          21, which was assumed to represent each day in January, February,
  vs. Always-Dark         6.7%     1.9%       14.4%   5.3%     1.1%        10.6%         March, November, and December. To consider the summer season, the
  vs. Rule-Based          0.0%     1.9%       9.9%    0.0%     1.1%        9.3%
                                                                                         representative day of June 21 was used to assess energy consumption

                                                                                   190
A. Faramarzi et al.                                                                                                                                         Solar Energy 218 (2021) 180–194

Table 6
Total daily source energy consumption for the small and medium size office buildings under different temperature threshold for Rule-Based Control for the days under
study.
  Total daily source energy consumption (J)

  Temperature              Small size office                                                                     Medium size office

                           Sep 30               Dec 21                June 21               March 21             Sep 30                   Dec 21           June 21            March 21
                                      8                   9                      9                     8                      9                    9                  10
  T = 11.6   ◦
               C           6.04 ×   10          1.38 × 10             1.56 ×   10           9.69 × 10            6.43   × 10              9.25 × 10        1.80   × 10        7.17   × 109
  T = 12.6   ◦
               C           6.02 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           6.42   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 13.6   ◦
               C           5.86 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           6.45   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 14.6   ◦
               C           5.86 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           6.48   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 15.6   ◦
               C           5.85 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           6.54   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 16.6   ◦
               C           5.87 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           6.77   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 17.6   ◦
               C           6.21 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           7.15   × 109             9.25 × 109       1.80   × 1010      7.17   × 109
  T = 18.6   ◦
               C           6.88 ×   108         1.38 × 109            1.56 ×   109          9.69 × 108           7.77   × 109             9.25 × 109       1.80   × 1010      7.17   × 109

                                                                                                     The total inflating/switching action cost for constant utility rate is
Table 7
                                                                                                     estimated as $581 per year for the small office building and $3,435 per
Total daily source energy consumption for medium size office buildings during
                                                                                                     year for the medium size office building. For the TOU program, the cost
the September simulation day for default and optimum threshold temperature of
                                                                                                     is highest in the shoulder season, followed by the summer and winter
Rule-Based Control.
                                                                                                     seasons. It is interesting to observe that the winter season, which was the
  Total daily source energy consumption, J
                                                                                                     highest costly season in the constant rate program, is the cheapest sea­
  Control          Rule-Based Control with default       Rule-Based Control with                     son in the TOU program. The reason is due to lower numbers of switches
  Strategy         threshold temperature of 15.6 ◦ C     optimum threshold temperature               in the winter along with the varying cost of utility rates in the TOU
                                                         of 12.6 ◦ C
                                                                                                     program that involves more off-peak usage. The total cost for inflating/
  Rule-            6.54 × 109                            6.42 × 109                                  switching under the TOU program is estimated as $336 per year for the
    Based
                                                                                                     small office building and $2,345 per year for the medium size office
                                                                                                     building. Therefore, based on these estimates, adoption of a TOU pro­
and costs of each day in June, July, and August.                                                     gram could yield approximately 42% and 31% in annual cost savings for
    For the inflating/switching cost estimation, we used the recorded                                ETFE switching in the small and medium size office buildings,
current data (Amps) of the compressors operating in the Kaplan Institute                             respectively.
building at Illinois Institute of Technology to calculate the power/en­
ergy consumption of the compressors for dark and bright status, using                                4.6.2. Cost of ETFE compared to conventional construction
                  √̅̅̅
the formula P = 3 × V × I × pf, where V is the voltage in Volts, I is the                                This section evaluates the total construction cost of switchable ETFE
current in Amps and pf is the power factor, which we assumed to be 0.85                              and its required equipment (e.g., compressor and piping), including the
for three phase compressors. We then calculated the required power                                   cost of purchase, labor for installation, and profit cost, and compares the
draw per square meter of ETFE envelope area (W/m2) for each switching                                cost with a conventional construction assembly, which is assumed to be
action between dark and bright status. Based on the optimal ETFE                                     a double-glazed curtain wall. The curtain wall is assumed to consist of
schedule that we estimated for each side of the building and the ETFE                                two 3 mm clear glass layers with a 13 mm air gap between them with
area for each of small and medium size office building, the power/en­                                total assembly U-value of 2.73 W/m2K, which is nearly the same U-value
ergy consumption of the compressors to change the ETFE status is                                     as the ETFE construction to provide a fair comparison. Each glass has
calculated for each representative day. This approach is simplistic and                              solar transmittance of 0.837, visible transmittance of 0.898, and thermal
does not use daily data beyond the four simulation days but provides                                 conductivity of 0.9 W/m-K. Table 11 presents assumptions for the cost of
some insight into the potential magnitudes of inflating/switching energy                             construction including materials, compressors and air dryer, and piping
costs throughout the year without having to run computationally                                      for the ETFE enclosure. These costs were obtained by averaging the
expensive annual simulations.                                                                        quoted costs from multiple original quotes for the ETFE enclosure on the
    Table 10 shows the estimated cost of inflating/switching the ETFE                                Kaplan Institute building, provided by IIT’s Facilities Department. Only
status using air compressor(s) for each of the shoulder, winter, and                                 material costs are shown for the assumed curtain wall assembly, which
summer seasons, as defined, for both the small and medium size office                                were estimated using RSMeans (RSMeans Online, n.d.) with the design
buildings based on both utility rate assumptions. As it is observed in                               documents for the Kaplan Institute building as a guide for sizing the
Table 10, for the constant utility rate, the cost for ETFE inflating/                                conventional enclosure. For example, for piping, we considered the size
switching is highest in winter, followed by the shoulder and summer                                  of the pipes, which were designed exclusively for the Kaplan Institute
seasons. The reason for this hierarchy is due in part to the number of                               building, and used that to evaluate the cost of piping assuming different
months that were considered for each season (i.e., 5 months for the                                  lengths for the small and medium size office buildings. We assumed four
winter season, 4 months for the shoulder season, and 3 months for the                                compressors with 7.457 kW (10 HP) for each façade cardinal direction
summer season), as well as differences in inflating/switching patterns.                              (i.e., north, east, south, ad west) of the medium size office building, and
                                                                                                     four compressors with 1.49 kW (2 HP) each for the small office building.

Table 8
                                                                                                     Table 9
Total daily source energy savings from using the Optimal Control strategy
                                                                                                     Utility rate of constant and TOU programs for the city of Chicago, IL.
compared to Rule-Based strategy with default and optimum threshold temper­
ature in the medium size office building during the September simulation day.                          Utility rate program       Off-Peak         Peak                      Super peak
                                                                                                                                  (10 pm-6         (6 am-2 pm & 7 pm-10      (2 pm-7 pm)
  Total daily source energy savings (%)                                                                                           am)              pm)
  Optimal Control savings compared to the Rule-Based             T=              T=                    Constant (₵/kWh)           5.847            5.847                     5.847
  control with different threshold temperatures                  15.6 ◦ C        12.6 ◦ C              Time-Of-Use                1.779            2.710                     12.867
  vs. Rule-Based                                                 3.7%            1.9%                    (₵/kWh)

                                                                                               191
You can also read